Differential Changes in the Cellular Composition of the Developing Marsupial Brain

Differential Changes in the Cellular Composition ofthe Developing Marsupial Brain

Adele M.H. Seelke,1 James C. Dooley,1 and Leah A. Krubitzer1,2*1Center for Neuroscience, University of California, Davis, Davis, California 956182Department of Psychology, University of California, Davis, Davis, California 95618

ABSTRACTThroughout development both the body and the brain

change at remarkable rates. Specifically, the number of

cells in the brain undergoes dramatic nonlinear

changes, first exponentially increasing in cell number

and then decreasing in cell number. Different cell types,

such as neurons and glia, undergo these changes at dif-

ferent stages of development. The current investigation

used the isotropic fractionator method to examine the

changes in cellular composition at multiple developmen-

tal milestones in the short-tailed opossum, Monodelphis

domestica. Here we report several novel findings con-

cerning marsupial brain development and organization.

First, during the later stages of neurogenesis (P18),

neurons make up most of the cells in the neocortex,

although the total number of neurons remains the same

throughout the life span. In contrast, in the subcortical

regions, the number of neurons decreases dramatically

after P18, and a converse relationship is observed for

nonneuronal cells. In the cerebellum, the total number

of cells gradually increases until P180 and then remains

constant, and then the number of neurons is consistent

across the developmental ages examined. For the three

major structures examined, neuronal density and the

percentage of neurons within a structure are highest

during neurogenesis and then decrease after this point.

Finally, the total number of neurons in the opossum

brain is relatively low compared with other small-

brained mammals such as mice. The relatively low num-

ber of neurons and correspondingly high number of

nonneurons suggests that in the marsupial brain non-

neurons may play a significant role in signal processing.

J. Comp. Neurol. 521:2602–2620, 2013.

VC 2013 Wiley Periodicals, Inc.

INDEXING TERMS: development; marsupials; neocortex; evolution; comparative neuroanatomy; isotropic fractionation

Throughout development, both the body and the brain

undergo remarkable changes in both size and function.

Because the nature of brain–body relationships and the cel-

lular composition and state of connectivity of the brain

itself are fundamentally different at various stages of devel-

opment, it is important to consider each developmental

stage in its own right rather than as a continuum of the

same structure that changes from simple to complex. This

is particularly true because some cells play radically differ-

ent roles at different stages of development (Polazzi and

Contestabile, 2002). Traditionally, when we consider brain

development, we focus on neural development: how neu-

rons are generated; how they migrate; and ultimately how

they differentiate, connect, and refine their structure. The

importance of neural development is undeniable, but the

brain is not composed of neurons alone. Other cell types,

including but not limited to endothelial cells (capillaries and

blood vessels), mesothelial cells (pia mater), ependymal

cells (lining of the ventricles), and glial cells, are present as

well, and, among these, glia are the most prevalent (Morest

and Silver, 2003; Temple, 2001). Recent studies have

underscored the importance of microglia cells in both neu-

rogenesis (Cunningham et al., 2012) and programmed cell

death (Kriegstein and Noctor, 2004; Polazzi and Contesta-

bile, 2002; Upender and Naegele, 1999) and of astrocytes

in synaptic transmission and plasticity in adults (Nadarajah

and Parnavelas, 2002; Rakic, 1990; Santello et al., 2012).

The current study examines the developmental rela-

tionships between neuronal and nonneuronal cells in the

brains of short-tailed opossums using the isotropic frac-

tionator technique. This relatively new methodology

allows one to estimate quickly and reliably the number of

neuronal and nonneuronal cells in different structures of

the brain (Herculano-Houzel and Lent, 2005). This is

Grant sponsor: National Institutes of Health; Grant number:R21NS071225 (to L.A.K.); Grant number: T32EY015387 (to J.C.D.).

*CORRESPONDENCE TO: Leah A. Krubitzer, Center for Neuroscience,1544 Newton Ct., Davis, CA 95616.. E-mail: [email protected]

VC 2013 Wiley Periodicals, Inc.

Received August 11, 2012; Revised September 18, 2012; AcceptedJanuary 4, 2013

DOI 10.1002/cne.23301

Published online January 16, 2013 in Wiley Online Library(wileyonlinelibrary.com)

2602 The Journal of Comparative Neurology | Research in Systems Neuroscience 521:2602–2620 (2013)

RESEARCH ARTICLE

accomplished by transforming anisotropic structures of

the brain, such as the six-layered neocortex, into an iso-

tropic suspension of cell nuclei. These nuclei can be la-

beled using immunohistochemical techniques that identify

them as neurons or other types of cells. The nuclei are

then counted, and the total number of cells, as well as the

number of neuronal and nonneuronal cells, can be calcu-

lated. Because these techniques are independent of total

brain volume, they are ideal for a number of important

developmental comparisons across different age groups in

which overall brain volume changes dramatically.

We used short-tailed opossums, because they are born

in a very immature state, the equivalent to embryonic day

(E) 13–14 in rats (Saunders et al., 1989). Furthermore,

infant opossums are not housed within a pouch, so they

are easily accessible for manipulation and tissue harvest.

Finally, compared with placental (eutherian) mammals

such as rodents, the significant events of brain and sen-

sory development occur over a protracted period, extend-

ing past the first postnatal month (Fig. 1; Saunders et al.,

1989). Thus, the short-tailed opossum is an excellent

model for studying both early brain development and

adult cortical composition.

The goal of the present investigation was to examine

the neuronal and nonneuronal composition of different

structures of the developing brain, including the neocor-

tex, cerebellum, and subcortical structures across impor-

tant developmental time points. We found that, whereas

the total number of cells increases across development,

the proportion of neurons generally decreases across de-

velopment, and the cellular composition of different

regions of the brain follows distinct and different develop-

mental trajectories. These findings represent the first

quantification of the cellular composition of a marsupial

brain across several important developmental stages.

MATERIALS AND METHODS

The isotropic fractionation procedure consists of

multiple stages. First, tissue was dissected into major

structures. Next, tissue was processed, which included

homogenization, DAPI staining, and NeuN immunohisto-

chemistry. The third stage involved quantifying the number

of DAPI-labeled nuclei within a sample, and the fourth stage

involved determining the proportion of NeuN-labeled nuclei

within that same sample. Finally, in the last stage, we used

these values to calculate the total number of cells, total

number of neurons, total number of nonneurons, cell

Figure 1. Timeline of significant postnatal developmental milestones in the life of Monodelphis domestica. Thalamocortical connections

are formed between P5 and P12 (red). Peak cortical neurogenesis occurs from P14 to P24 (yellow). Eyes open between P31 and P34

(blue). Opossums are separated from their mothers at P56 (gray) and reach sexual maturity by P180 (green). Their peak reproductive pe-

riod is from P270 to P365, and after P365 they enter old age.

TABLE 1.

Age, Sex, and Weight of Animals

Case No. Group Weight (g) Sex

12-060 P18 3 f12-061 P18 3 m12-062 P18 2 m12-064 P18 2 f

11-148 P35 7 f11-149 P35 9 f11-235 P35 7 f11-236 P35 5 f11-237 P35 7 m

11-157 P56 13 f11-158 P56 13 f11-160 P56 15 f11-240 P56 13 m11-241 P56 13 m

11-150 P180 77 m11-159 P180 96 f11-162 P180 59 f11-267 P180 99 m11-268 P180 131 m

11-154 P270–365 75 f11-156 P270–365 137 m11-161 P270–365 64 f11-242 P270–365 86 f

11-151 >P365 148 m11-152 >P365 121 f11-153 >P365 107 f11-155 >P365 134 m11-243 >P365 109 m

Opossum brain composition across development

The Journal of Comparative Neurology | Research in Systems Neuroscience 2603

density, neuronal density, and nonneuronal density. Each

of these stages is discussed in detail below.

SubjectsThirty South American short-tailed opossums (Mono-

delphis domestica) raised in our breeding colony at the

University of California, Davis, were used in these experi-

ments. Opossums were examined at six developmentally

significant ages: during cortical neurogenesis (P18, N ¼4), eye opening (P35 6 1 day, N ¼ 5), weaning (P56 6 1

day, N ¼ 5), sexual maturity (P180 6 1 day, N ¼ 5),

adulthood (P270–365, N ¼ 4), and elderly adulthood

(>P365, N ¼ 5; see Table 1 for sexes and weights). Two

additional animals, one at P18 and one at P180, were

used to show the relationship between neurons and non-

neurons in nonhomogenized tissue. All experiments were

performed under National Institutes of Health guidelines

for the care of animals in research and were approved by

the Institutional Animal Care and Use Committee of the

University of California, Davis.

Tissue dissectionAnimals were euthanized with an overdose of sodium

pentobarbital (Beuthanasia; 250 mg/kg) and transcar-

dially perfused with phosphate-buffered saline, followed

by 4% paraformaldehyde. The brain was extracted, photo-

graphed, and weighed. Under a microscope, the brain

was dissected into sections: the left cortical hemisphere,

right cortical hemisphere, cerebellum, and remaining sub-

cortical regions (including the hypothalamus, thalamus,

and brainstem; Fig. 2). The neocortex was then isolated

by removing the hippocampus, basal ganglia, pyriform

cortex, and olfactory bulb from the cortical hemisphere.

These sections were photographed and weighed, then

placed in 5% paraformaldehyde for storage.

Tissue processingTissue was homogenized in a 15-ml glass Kontes Ten-

broek tissue grinder (Kimble Chase) with a dissociation

solution composed of 10 ml Triton X-100 and 11.76 g so-

dium citrate in 1,000 ml distilled water. The homogeniza-

tion process broke down cell membranes, producing a

suspension of isolated cellular nuclei with no visible tis-

sue clumps. An aliquot from the main suspension was

centrifuged and resuspended in a solution of phosphate-

buffered saline (PBS) and 40,6-diamidino-2-phenylindole

(DAPI), which binds to DNA and labels all cellular nuclei

regardless of cell type (Fig. 3A,D,G). If the tissue had

been stored in fixative for more than 4 weeks, the sample

was suspended in a boric acid solution (12.37 g in 1,000

ml distilled water) and placed in an oven at 70�C for 1

hour for epitope retrieval. A separate aliquot from the

main suspension was stained for neuronal nuclei using

immunocytochemical techniques with the anti-NeuN anti-

body (Millipore, Bedford, MA; see Table 2 for antibody

description). Alexa Fluor 647 or Alexa Fluor 700 goat

anti-mouse IgG secondary antibody (Invitrogen, Carlsbad,

CA) was used to fluorescently tag NeuN-labeled nuclei

(Fig. 3B,E,H). In selected cases, we also used Alexa Fluor

594 goat anti-mouse IgG secondary antibody (Invitrogen)

to tag NeuN-labeled nuclei fluorescently.

Antibody characterizationSee Table 2 for a list of the antibodies used. NeuN anti-

body (clone A60) specifically recognizes the DNA-binding,

neuron-specific protein NeuN, which is present in most

Figure 2. Dissection of the brain for isotropic fractionation. The whole brain (A) is separated into the left and right cerebral hemispheres,

subcortical structures (including the midbrain, thalamus, hypothalamus, and brainstem), and cerebellum (B). The cerebral hemispheres are

further dissected, removing the olfactory bulb, pyriform cortex, basal ganglia, and hippocampus, until only the neocortex remains (C). Scale

bars ¼ 5 mm in A,B; 1 mm in C.

A.M.H. Seelke et al.

2604 The Journal of Comparative Neurology |Research in Systems Neuroscience

central and peripheral nervous system neuronal cell types

of all vertebrates tested. However, there are some excep-

tions that are described in the Discussion. In a Western

blot analysis, this antibody recognized two or three bands

in the 46–48 kDa range and possibly another band at

approximately 66 kDa (adapted from product informa-

tion). Because NeuN specifically stains neuronal nuclei,

staining of nonneural tissue was used as a negative

control.

Nuclei quantificationThe total number of nuclei was determined using a

Neubauer cell-counting chamber (Optik Labor). Samples

were vortexed, and 10 ll aliquots were immediately

loaded into a Neubauer cell-counting chamber and placed

on a fluorescence microscope for visualization and count-

ing of nuclei. Standard stereological protocols were used

(Mouton, 2002). To obtain a reliable and representative

sample of the number of DAPI-labeled nuclei within a

Figure 3. Distribution of DAPIþ and NeuNþ nuclei following tissue dissociation. Cellular nuclei from the neocortex (top row), subcortical

regions (middle row), and cerebellum (bottom row) were dissociated and stained for DAPI (labeled in blue; A,D,G) and NeuN (labeled in

red; B,E,H). Nuclei that are stained for both DAPI and NeuN are considered to be neuronal (labeled in pink; C,F,I). Arrows identify one nu-

cleus in each brain region that is both DAPIþ and NeuNþ. Color and contrast have been adjusted in Adobe Photoshop. Scale bars ¼ 50

lm.

TABLE 2.

Antibody Used

Antigen Immunogen Maunfacturer, species, mono- vs. polyclonal, catalog No. Dilution used

NeuN Purified cell nuclei from mouse brain Millipore (Billerca, MA), mouse monoclonal, MAB377 1:300



given sample size, we counted the nuclei in one 8-square

by 8-square section of the Neubauer chamber (for details

see Campi et al., 2011). We repeated these counts on

multiple samples until we had counted 10 unique 8 � 8

sections. We then calculated the mean number of nuclei

in an 8 � 8 section and used that number to determine

the total number of nuclei in the sample, using the equa-

tion described below.

Determining NeuN1 percentageWith a separate aliquot from the main suspension, the

ratio of neuronal to nonneuronal nuclei within a sample

was determined using a flow cytometer at the UC Davis

Flow Cytometry Shared Resource Center. The use of a

flow cytometer to automate the detection and counting

of neuronal and nonneuronal nuclei is both faster and

more reliable than visual inspection under a fluorescent

microscope (Collins et al., 2010b). To quantify the propor-

tion of neuronal nuclei, we used a Becton Dickinson (BD)

Five-Laser LSRII flow cytometer. The violet laser (50 mW,

405 nm) excites DAPI-labeled nuclei, and the red laser

(50 mW, 635 nm) excites both Alexa Fluor (AF) 647-la-

beled nuclei and AF 700-labeled nuclei. For all samples,

between 1,000 and 10,000 DAPI-positive nuclei were

evaluated for AF 647 and AF 700 label. Samples run on

the flow cytometer were forced through a 35-lm mesh

cell filter, vortexed, and immediately taken into the LSRII.

Selection gates were determined by a flow cytometry

expert who was blind to the developmental stage and

brain areas of the samples. The selection gates used to

estimate NeuN-IR nuclei were constructed to minimize

the amount of both myelin debris and clumps of nuclei

(i.e., doublets and triplets) that were counted within the

samples. The samples were run for 120 seconds or until

10,000 nuclei had been detected. The ratio of neuronal to

nonneuronal nuclei was estimated from the gated popula-

tion (see equations below).

To ensure the reliability of our results, in several cases

we reanalyzed samples of dissociated and stained tissue.

We did not see any significant differences between the

first and second sets of analyses. Additionally, as men-

tioned above, in selected cases we labeled neuronal

nuclei with AF 594 and counterstained them with DAPI.

This allowed us to perform manual counts of NeuNþ and

DAPIþ nuclei on a fluorescent microscope. The UV filter

excited the DAPI-labeled nuclei, and the TRITC filter

excited the AF 594-labeled nuclei. We used a Neubauer

cell counting chamber to count the number of DAPI-la-

beled and AF 594-labeled nuclei. We then calculated the

proportion of NeuNþ to DAPIþ nuclei and compared that

with the results obtained using the flow cytometer. We

did not see any significant differences between these

sets of analyses.

EquationsEstimates of cellular composition for a given structure

were derived from the following equations.

Total nuclei ¼ ½ðnumber of DAPIþ nucleiÞ=ðvolume of suspension counted in mm3Þ�� ðtotal suspension volume in cm3Þ � 1; 000

Percentage neurons ¼ ðnumber of NeuNþ nucleiÞ=ðnumber of DAPIþ nucleiÞ

Total neurons ¼ percent neurons � total nuclei

Total nonneurons ¼ total nuclei� total neurons

Cell density ¼ number of cells=weight of structure

Neuron density ¼ number of neurons=weight of structure

Nonneuron density ¼ number of nonneurons

=weight of structure

AnalysisDevelopmental changes in the weights of the whole

brain, neocortex, subcortical structures, cerebellum, and

brain weight to body weight ratio were assessed by

ANOVA (JMP; SAS, Cary, NC), and differences between

specific age groups were determined using Student’s t-

tests. Likewise, developmental differences in the percent-

age of neurons, total number of cells, total number of

neurons, total number of nonneurons, cell density, neuro-

nal density, and nonneuronal density were assessed by

ANOVA, and differences between specific age groups

were determined by using Student’s t-tests. For all tests,

a¼ 0.05.

Whole-section stainingIn two additional cases, we show the relationship

between DAPI- and NeuN-labeled cells in nonhomogen-

ized tissue (Figs. 4–6). Each brain was sliced on a freezing

microtome into 50-lm horizontal sections. Briefly, the 50-

lm-thick free-floating sections were first rinsed (3 � 5

minutes) in PBS. To quench endogenous peroxidase, sec-

tions were incubated in an aqueous solution of 10%

MeOH and 3% H2O2 for 30 minutes at room temperature.

After rinses in PBS-0.1% Triton X-100 (3 � 10 minutes),

nonspecific binding was suppressed by a preincubation in

10% normal goat serum (NGS; Invitrogen) and PBS-0.1%

Triton X-100 for 1 hour at room temperature. Sections

were then transferred to the primary antibody solution

(mouse anti-NeuN, 1:100; Millipore) containing 10% NGS

and PBS-0.1% Triton X-100 overnight at 4�C. Tissue sec-

tions were then rinsed in PBS-0.1% Triton X-100 (4 � 10

minutes) and incubated in a secondary antibody solution

(Cy3 goat anti-mouse IgG, 1:200; Millipore) for 4 hours at

room temperature. The tissue sections were then



thoroughly rinsed in PBS (3 � 10 minutes). The fluores-

cent sections were double-labeled with the nuclear stain

DAPI (Figs. 4–6), and sections were rinsed in PBS (3 � 5

minutes), mounted on gelatin-subbed slides, and cover-

slipped. Adjacent tissue sections were stained for Nissl

so that laminar patterns visualized with Nissl stains could

be compared with DAPI and NeuN labeling (Fig. 6).

RESULTS

The weight of the whole brain (including both cortical

hemispheres, thalamus, hypothalamus, cerebellum, and

brainstem; shown at each age in Fig. 7) significantly

increased across development (F5,27 ¼ 111.88,

P< 0.0001; see Table 3 for values). Likewise, the weights

of the neocortex (F5,27 ¼ 47.05, P < 0.0001), subcortical

regions (F5,27 ¼ 110.47, P < 0.0001), and cerebellum

(F4,23 ¼ 40.87, P < 0.0001) also significantly increased

across development (Fig. 8A). In contrast, the brain

weight/body weight ratio significantly decreased across

development (F5,27 ¼ 69.09, P < 0.0001; Fig. 8B), indi-

cating that, although both the body and the brain

increased in weight across the life span, the body grew at

a much faster rate than the brain.

Different regions of the brain showed differential pat-

terns of growth across development (Fig. 8C). At each

age, the proportion of the brain comprising the neocortex,

subcortical regions, and cerebellum was determined by

dividing the weight of the structure in question by the

weight of the whole brain. The proportion of the brain

consisting of the neocortex was 7.7% at P18, increasing

to 10.2% at P35 and significantly decreasing to 8.0% at

P180 and remaining at that level throughout adulthood

(F5,27 ¼ 12.82, P < 0.0001; Fig. 8C, Table 3). The propor-

tion of the brain comprising subcortical regions was high-

est at P18 at 44.0% and by P56 had significantly

Figure 4. The normal, anisotropic distribution of DAPIþ and NeuNþ cells in portions of the neocortex (A–C), dorsal thalamus (D–F), and

cerebellum (G–I) in a P180 brain. The superimposition of DAPIþ (blue; A,D,G) and NeuNþ (red; B,E,H) cells reveals the distribution of neu-

rons (pink; C,F,I) relative to all cells. In this low-power image the densely packed neurons in cortical layers 2–3 are clearly evident (C).

Neurons in this portion of the dorsal thalamus are more evenly distributed. The high-power images of the cerebellum (G–I) show that the

number of neurons is relatively low (H) compared with the total numbers of cells (G,I). Color and contrast have been adjusted in Adobe

Photoshop. Scale bars ¼ 100 lm.



decreased to 31.0%, where it remained through >P365

(F5,27 ¼ 67.93, P< 0.0001). The cerebellum, on the other

hand, significantly increased from 5.5% at P18 to 12.4%

at P56 (F5,27 ¼ 26.48, P < 0.05). These data indicate

that, although the brain grows steadily during the early

part of the life span, it does not grow uniformly.

To demonstrate the relationship between DAPI and

NeuN stains and the efficacy of NeuN in labeling neurons

in both very young and adult brains, the distribution of

cells and neurons was visualized by staining whole sec-

tions of P18 and P180 tissue (Figs. 4–6). All cells were

labeled using DAPI (Fig. 4A,D,G), and the tissue was coun-

terstained for NeuN to identify neurons (Fig. 4B,E,H). The

superimposition of the DAPIþ and NeuNþ images in

Figure 4C,F,I allowed us to visualize individual nuclei and

neurons and distinguish them from nonneuronal cells.

The relationship between DAPI and NeuN also was dem-

onstrated in the P18 opossum (Fig. 5), and for the neocor-

tex DAPI and NeuN labeling were directly compared with

adjacent Nissl-stained sections (Fig. 6). Both the P18 and

the P180 tissue showed a laminar distribution of neurons

in the cortex and cerebellum. In the developing neocor-

tex, laminar patterns, as revealed with both Nissl and

DAPI stains, indicated an abundance of cells in the ven-

tricular zone and subventricular zones, although few of

these cells were NeuN positive (Fig. 6A–C). Some NeuN-

positive cells were labeled in the subplate, and NeuN-

positive cells were found in abundance in the cortical

plate. In the adult, Nissl- and DAPI-stained sections

appeared similar, showing a similar laminar distribution

with a prominent layer 2–3 (Fig. 6D,E), and NeuN

revealed an abundance of neurons in these same layers

Figure 5. The normal, anisotropic distribution of DAPIþ and NeuNþ cells in portions of the neocortex (A–C), thalamus (D–F), and cerebel-

lum (G–I) in the brain of a P18 opossum. The superimposition of DAPIþ (blue; A,D,G) and NeuNþ (red; B,E,H) cells reveals the distribution

of neurons (pink; C,F,I). In these images, the laminar structures of the cortex and cerebellum are clearly identifiable in both the DAPIþ

and the NeuNþ cells. Within the thalamic section, there is a dense band of DAPIþ cells surrounding the ventricle. This band can also be

seen in the NeuNþ cells, but it is very faint, indicating that this area contains a large proportion of nonneuronal cells or premitotic cells.

Color and contrast have been adjusted in Adobe Photoshop. Scale bars ¼ 50 lm.



(Fig. 6D,E). Moderate numbers of neurons were observed

in deeper cortical layers (Fig. 6F).

The isotropic fractionator method utilizes these same

fluorescent stains and follows the same principle of

identifying neurons within a population of cells, but it

homogenizes the tissue, making different structures of

the brain (e.g., neocortex, cerebellum) isotropic. In this

way, the cellular composition of large pieces of neural

Figure 6. The normal, anisotropic distribution of cells in cortical tissue stained for Nissl (A,D), DAPI (B,E), and NeuN (C,F) in P18 (A–C)

and P180 (D–F) opossums. Both the P18 and P180 cortical sections are from a similar location in somatosensory cortex. At P18 the lami-

nar organization of the developing neocortex is distinct, and neurons within the cortical plate are clearly labeled with NeuN. Although the

ventricular zone is cell dense, it contains no NeuN-labeled cells. In the adult the characteristic layers of the neocortex are visible in both

Nissl- and DAPI-stained tissue. Tissue stained for NeuN indicates a lack of neurons in layer 1, dense labeling of neurons in layers 2–3,

and moderate labeling of neurons in layer 4 and 6. mz, Marginal zone; cp, cortical plate; sp, subplate; iz, intermediate zone; svz, subven-

tricular zone; vz, ventricular zone. Laminar divisions of early postnatal animals have been described by Cheung et al. (2010) and Saunders

et al. (1989). Color and contrast have been adjusted in Adobe Photoshop. Scale bars ¼ 100 lm.



tissue can be quantified. It should be noted, however,

that during the course of the homogenization areal and

laminar information is necessarily lost. After the homoge-

nization of the tissue and nuclear dissociation, all nuclei

were first labeled with DAPI, then counterstained for

NeuN in order to reveal neuronal nuclei (Fig. 3). In Figures

3–6 nonneuronal nuclei are visualized as blue, neuronal

nuclei are visualized as red, and nuclei that stain for both

DAPI and NeuN are visualized as pink.

The cellular composition of the neocortex changed sig-

nificantly across development (Fig. 9). The total number

of cells in the neocortex significantly increased across de-

velopment (F5,27 ¼ 7.19, P < 0.001; Fig. 9A). Although

the total number of neurons did not significantly change

(F5,27 ¼ 0.27, NS), the total number of nonneurons signifi-

cantly increased (F5,27 ¼ 15.68, P < 0.0001). Further-

more, the overall cell density decreased across develop-

ment (F5,27 ¼ 72.31, P < 0.0001; Fig. 9B), as did

Figure 7. Relative size of Monodelphis domestica brains at P18, P35, P56, P180, P270–365, and >P365. The size and shape of the brain

change dramatically during the first 6 months of life. The caudolateral expansion of the cerebral cortex is particularly apparent. After sex-

ual maturity (P180), there is little change in the size of the entire brain or the relative size of different structures. Images were converted

to gray scale and contrast and brightness levels were adjusted in Adobe Photoshop. Scale bars ¼ 2 mm.

TABLE 3.

Weights of Brain Regions

Age Whole brain Neocortex

Subcortical

regions Cerebellum

P18 0.1191 6 0.0026 0.0092 6 0.0004 0.0525 6 0.0014 0.0066 6 0.0009P35 0.3450 6 0.0176 0.0353 6 0.0031 0.1401 6 0.0055 0.0375 6 0.0042P56 0.5112 6 0.0062 0.0498 6 0.0024 0.1586 6 0.0060 0.0498 6 0.0011P180 0.8949 6 0.0568 0.0719 6 0.0057 0.2767 6 0.0142 0.1174 6 0.0085P270–365 0.8533 6 0.0404 0.0685 6 0.0038 0.2705 6 0.0142 0.1148 6 0.0068>P365 0.9833 6 0.0272 0.0726 6 0.0029 0.3096 6 0.0072 0.1235 6 0.0070

Age

Brain weight/

body weight

P18 0.0522 6 0.0087P35 0.0503 6 0.0031P56 0.0382 6 0.0011P180 0.0102 6 0.0010P270–365 0.0100 6 0.0011>P365 0.0080 6 0.0004

Age % Neocortex % Subcortical regions % Cerebellum

P18 7.72 6 0.33 44.05 6 0.36 5.52 6 0.17P35 10.18 6 0.44 40.73 6 1.06 10.75 6 0.80P56 9.73 6 0.41 31.00 6 0.86 12.40 6 0.21P180 8.01 6 0.22 31.03 6 0.52 13.15 6 0.55P270–365 8.03 6 0.27 31.69 6 0.43 13.44 6 0.39>P365 7.26 6 0.29 31.51 6 0.36 12.57 6 0.66



neuronal density (F5,27 ¼ 23.13, P < 0.0001) and non-

neuronal density (F5,27 ¼ 6.50, P < 0.001). The neuron

percentage (i.e., the percentage of total nuclei that were

positive for NeuN) significantly decreased across devel-

opment (F5,27 ¼ 3.10, P < 0.05; Fig. 9C; Table 4).

For subcortical regions, the cellular composition also

significantly changed across development (Fig. 10). The

total number of cells significantly increased with age

(F5,27 ¼ 20.06, P < 0.0001; Fig. 10A; Table 4). Neuronal

number was greatest at P18, during neurogenesis, and

then significantly decreased with age (F5,27 ¼ 3.48, P <

0.05), whereas the total number of nonneurons signifi-

cantly increased over the same period (F5,27 ¼ 43.62,

P < 0.0001). The overall cell density significantly

decreased across development (F5,27 ¼ 5.13, P < 0.005;

Fig. 10B), as did neuronal density (F5,27 ¼ 74.01, P <

0.0001), although the density of nonneurons increased

(F5,27 ¼ 27.51, P < 0.0001). The percentage of neurons

significantly decreased throughout the life span (F5,27 ¼157.73, P < 0.0001; Fig. 10C).

The cerebellum also showed significant changes in its

cellular composition across development (Fig. 11). The

total number of cells sharply and significantly increased

until P180, and then decreased through >P365 (F5,23 ¼30.43, P < 0.0001; Fig. 11A, Table 4). The total number

of neurons did not significantly change across

Figure 8. Changes in brain weight (A), brain/body ratios (B), and relative size of major structures (C) across development. The entire

brain and major structures increase in size until P180 and then remain relatively constant (A). Brain and body weight ratios decrease with

age and then stabilize at P180, indicating that the body grows at a much faster rate than the brain until sexual maturity (B). The major

structures that compose the brain undergo different trajectories of growth (C). Mean 6 SE. Values with different letters are significantly

different.



development (F5,23 ¼ 2.45, P ¼ 0.07), but the total num-

ber of nonneurons significantly increased until P180, at

which point it remained steady (F5,23 ¼ 44.36, P <

0.0001). Cell density increased from P18 through P35

and then gradually and significantly decreased with age

(F5,23 ¼ 5.43, P < 0.005; Fig. 11B). In contrast, neuron

density significantly decreased across development (F5,23¼ 6.46, P < 0.005), and the nonneuron density increased

from P18 to P35, then decreased from P35 to >P365

(F5,26 ¼ 15.26, P < 0.0001). Finally, similar to both the

neocortex and the subcortical regions, the percentage of

neurons within the cerebellum significantly decreased

with age (F5,27 ¼ 23.72, P < 0.0001; Fig. 11C).

DISCUSSION

The current study showed that the size of the body and

brain increased with age, and the ratio between the brain

and the body was relatively high early in development,

but both dropped at 6 months of age and remained con-

stant. In terms of cellular composition, the total number

of cells in the neocortex, subcortical structures, and cere-

bellum increased with age until 6 months. However, there

were some important differences in the growth patterns

and cellular composition across the major structures, par-

ticularly with regard to neuronal number. For example, in

the neocortex, the cellular density was highest at P18,

dropped off at P35, and remained constant across pro-

gressively older age groups. The number of neurons

remained relatively constant, whereas the percentage of

neurons declined with age (Figs. 9, 12) suggesting that

the growth of the neocortex is due to the addition of non-

neuronal cells and that naturally occurring neuronal cell

death occurs before P18 or is not prevalent in the devel-

oping marsupial neocortex. The increase in nonneuronal

cells is not surprising, in that our earliest sampled age

(P18) was during peak cortical neurogenesis, which is

then followed by gliogenesis within the neocortex

(Cheung et al., 2010; Miller and Gauthier, 2007; Puzzolo

and Mallamaci, 2010). In subcortical regions, neuron

number was high at P18, decreased at P35, and remained

constant at progressively older ages, as did the percent-

age of neurons; a converse relationship was observed for

nonneuronal cells. Neuronal density and percentage neu-

rons also dropped significantly at P35 compared with

P18, and a converse relationship was observed for non-

neuronal cells. Finally, in the cerebellum, cell number

increased with age and was accounted for mainly by an

increase of nonneuronal cells. As in the cortex, neuronal

number remained constant across all ages. The density

and percentage of neurons decreased dramatically after

P18, and a converse relationship in cell density was

observed for nonneuronal cells.

Figure 9. Changes in cellular composition of the neocortex. (A)

Changes in the total number of cells (left), number of neurons

(center), and number of nonneurons (right) in the neocortex at

different developmental stages. Although the total number of cells

and nonneurons increases across the ages we sampled, the num-

ber of neurons does not change. (B) Changes in total cell density

(left), neuronal density (center), and nonneuronal density (right)

across development. The density of both cell types was highest

at P18, decreased by P35, and then remained unchanged through

adulthood. (C) The percentage of all cells within the neocortex

that were neurons decreased across development. Mean 6 SE.

Values with different letters are significantly different.



TABLE 4.

Cellular Composition of Brain Regions

Age % Neurons

No. cells

(in millions)

No. neurons

(in millions)

No. nonneurons

(in millions)

No. cells/gram

(in millions)

No. neurons/gram

(in millions)

No. nonneurons/gram

(in millions)

NeocortexP18 60.55 6 9.44 1.637 6 0.058 .985 6 0.139 .653 6 0.163 180.032 6 13.690 109.054 6 18.773 70.977 6 18.490P35 44.44 6 9.21 1.580 6 0.141 .743 6 0.093 .836 6 0.093 47.043 6 6.952 22.792 6 5.787 24.250 6 3.103P56 43.52 6 11.69 2.027 6 0.220 .970 6 0.131 1.057 6 0.131 40.454 6 3.142 19.059 6 6.024 21.394 6 2.928P180 22.04 6 3.63 3.224 6 0.440 .748 6 0.283 2.476 6 0.283 44.578 6 4.136 10.111 6 2.132 34.468 6 2.678P270–365 33.55 6 3.71 2.606 6 0.308 .847 6 0.050 1.760 6 0.306 37.901 6 3.146 12.509 6 1.226 25.391 6 3.297>P365 25.62 6 4.86 3.030 6 0.209 .803 6 0.199 2.227 6 0.120 42.067 6 2.623 10.917 6 2.545 30.779 6 1.675Subcortical regionsP18 77.40 6 5.00 3.184 6 0.185 2.486 6 0.267 .698 6 0.117 60.693 6 3.293 47.436 6 5.288 13.257 6 2.198P35 6.77 6 2.67 7.187 6 0.659 .503 6 0.199 6.684 6 0.610 51.355 6 4.339 3.789 6 1.677 47.565 6 3.351P56 6.17 6 0.69 8.240 6 0.287 .512 6 0.066 7.728 6 0.252 52.090 6 1.759 3.186 6 0.286 48.904 6 1.868P180 6.30 6 0.99 12.496 6 1.009 .803 6 0.151 11.693 6 0.912 44.941 6 1.637 2.868 6 0.498 42.073 6 1.344P270–365 4.55 6 0.32 12.073 6 1.043 .556 6 0.079 11.517 6 0.971 44.636 6 3.108 2.053 6 0.277 42.582 6 2.853>P365 7.25 6 0.90 13.147 6 6.223 .959 6 0.172 12.044 6 0.632 42.577 6 2.326 3.079 6 0.510 38.885 6 2.541CerebellumP18 86.93 6 4.55 1.613 6 0.135 1.392 6 0.103 .221 6 0.084 245.857 6 21.530 212.595 6 17.028 33.262 6 12.790P35 19.31 6 8.14 14.874 6 0.635 2.740 6 1.051 12.134 6 1.521 416.778 6 46.719 90.409 6 49.516 326.368 6 33.137P56 30.82 6 5.67 24.155 6 1.896 5.794 6 1.732 18.36 6 1.039 374.539 6 31.199 90.037 6 27.103 284.501 6 5.000P180 12.75 6 4.20 36.835 6 3.654 5.122 6 1.889 31.713 6 2.399 314.390 6 22.527 43.320 6 16.131 271.070 6 12.177P270–365 9.97 6 6.32 32.893 6 2.282 1.258 6 0.401 31.363 6 1.914 279.857 6 21.313 10.702 6 3.633 269.154 6 17.996>P365 8.45 6 4.5 30.348 6 2.903 1.275 6 0.342 29.073 6 2.650 239.780 6 17.196 9.953 6 2.703 229.827 6 15.117

Opossu

mbrain

compositio

nacro

ssdevelopment

TheJournalofComparative

Neurology|Research

inSystem

sNeuroscience

2613

It should be noted that NeuN is a marker for postmitotic

neurons (Mullen et al., 1992; Sarnat et al., 1998). That is,

NeuN is not expressed in neuronal precursor cells but is

expressed as the neurons exit the cell cycle (Martinez-Cer-

deno et al., 2012; Noctor et al., 2008; Oomman et al.,

2004; Yan et al., 2001). Thus, the large proportion of

NeuN-positive cells during early development was not a

result of labeling blastocysts or other immature cell types.

Several important caveats must be considered when

interpreting data that use NeuN as a neuronal marker in

different tissue and in different species. First, NeuN fails to

label several types of neurons in the adult brain, such as

mitral cells in the olfactory bulb, retinal photoreceptors,

and Purkinje cells in the cerebellum (Mullen et al., 1992).

Second, NeuN fails to label some groups of postmitotic

neurons, such as layer VIa cells in the neocortex (Lyck

et al., 2007) and nongranule cell interneurons in the

mouse cerebellum (Weyer and Schilling, 2003) until later

developmental ages. The latter investigation also indicates

that the expression of NeuN during development may be

dependent on the physiological status of the developing

neurons (Weyer and Schilling, 2003). Thus, studies that uti-

lize NeuN to examine patterns of cellular composition

across multiple developmental time points must be inter-

preted with caution. Finally, although there is good evi-

dence that NeuN labels neurons in adult mammalian nerv-

ous tissue, and the use of the isotropic fractionator

methodology in over 30 species, including capybaras, star-

nosed moles, bonnet macaques, and baboons, is based on

this assumption (see, e.g., Burish et al., 2010; Collins

et al., 2010a; Herculano-Houzel et al., 2006; Sarko et al.,

2009), it is likely that species-specific differences in its

labeling pattern exist. For example, NeuN does not label

substantia nigra neurons in the gerbil but does label these

neurons in rats (Kumar and Buckmaster, 2007). Thus, com-

parative studies on specific details of its labeling patterns

for species other than mice and rats are important for

accurate interpretation of data from this methodology.

An example of how these issues impact our own inves-

tigation comes from our estimates of neuronal and non-

neuronal populations in the cerebellum where the num-

bers are low compared with other species. As noted

above, NeuN does not label Purkinje cells in the adult cer-

ebellum and may not label all interneurons in the cerebel-

lum or layer 6 cortical neurons until later developmental

Figure 10

Figure 10. Changes in cellular composition of subcortical

regions. (A) Changes in the total number of cells (left), number of

neurons (center), and number of nonneurons (right) in the neocor-

tex at different developmental stages. The total number of cells

and nonneurons increased across development, but the number

of neurons was highest at P18, decreased by P35, and remained

unchanged across development. (B) Changes in the total cell den-

sity (left), neuronal density (center), and nonneuronal density

(right) across development. The total cell density and neuronal

density were highest at P18. Although the total cell density grad-

ually decreased across development, neuronal density decreased

dramatically by P35 and thereafter remained constant. In con-

trast, nonneuronal density was lowest at P18, increased by P35,

then decreased slightly throughout adulthood. (C) The percentage

of neurons was highest at P18, significantly decreased by P35,

and remained constant across development. Mean 6 SE. Values

with different letters are significantly different.



ages. If this is the case for Monodelphis, then, at earlier

development ages, neurons in these structures may be

underrepresented. However, the low numbers of cells in

the developing cortex and cerebellum in the developing

Monodelphis are also observed in adults, suggesting that

there may be true species differences in marsupials and

small-brained eutherian mammals. These differences are

discussed below.

Neural development in marsupials androdents

Although this is the first study in marsupials to examine

and quantify the cellular composition across major neural

structures through different developmental time points,

there are other studies of neural development in marsu-

pials, particularly in the neocortex. As in other mammals,

neurogenesis in the marsupial neocortex occurs in a ros-

trolateral to mediocaudal progression (Molnar et al., 1998;

Sanderson and Weller, 1990). In marsupials, this process

is prolonged and occurs almost completely postnatally,

and, in some marsupials, such as the native cat, brush-

tailed possum, and wallaby, it occurs over a 2–3-month

postnatal period (Aitkin et al., 1991; Marotte and Sheng,

2000; Sanderson and Weller, 1990). Neurogenesis and

gliogenesis have been examined specifically in Monodel-

phis as well, and their duration may be somewhat shorter

than in Australian marsupials (Puzzolo and Mallamaci,

2010). From bromodeoxyuridine (BrdU) pulse-chase birth-

dating analysis, Puzzolo and Mallamaci suggest that neuro-

genesis was complete by P16. Neurons born after P18

remain mostly beneath the cortical plate; ages past P18

were not examined. By P30 cells born at P16 have

migrated to the superficial layers of the neocortex, indicat-

ing that the laminar development of the neocortex is com-

plete. It should be noted that the samples in this previous

study were taken from midfrontal cortex, where the wave

of neurogenesis begins and ends earlier than in other por-

tions of the neocortex. Other studies examining develop-

ment of the neocortex inMonodelphis indicate that cortical

neurogenesis occurs over a longer postnatal period; that

the characteristic developmental layers, including the ven-

tricular zone and subventricular zone, are still clearly appa-

rent at P45 (Saunders et al., 1989); and that late-stage

neurogenesis occurs in the middle of the fourth postnatal

week (Molnar et al., 1998). Our own data indicate that

peak neurogenesis of the neocortex and subcortical struc-

tures is complete by P35, because neuronal number is rel-

atively constant across ages sampled for the neocortex

(P18 to adulthood), and decreases at P35 for subcortical

structures. Our studies also indicate that gliogenesis is

prolonged in all structures of the brain that were examined

and extends into the sixth postnatal month of life (P180).

Figure 11. Changes in cellular composition of the cerebellum. (A)

Changes in the total number of cells (left), number of neurons (center),

and number of nonneurons (right) in the neocortex at different devel-

opmental stages. The total number of cells significantly increased

from P18 through P180, then decreased through>P365. The number

of neurons did not significantly change across development. In con-

trast, the number of nonneurons increased from P18 through P180,

then remained constant. (B) Changes in the total cell density (left),

neuronal density (center), and nonneuronal density (right) across de-

velopment. Both total cell density and nonneuronal density increased

from P18 to P35, then decreased through >P365. In contrast, neuro-

nal density was highest at P18, then decreased through adulthood.

(C) The percentage of neurons decreased across the life span. Mean

6 SE. Values with different letters are significantly different.



There are only two other studies in which cellular com-

position of the brain has been examined across develop-

mental age groups: one in mice (Lyck et al., 2007) and

one in rats (Bandeira et al., 2009). The former study only

examined the neocortex. However, neither of these stud-

ies captured early embryonic stages, for which it has

Figure 12. Diagram of changes in cellular composition across development. The major assumption here is that cells (neuronal and non-

neuronal) occupy the entire structure and that cell size does not change dramatically. This assumption is purposely simplified to empha-

size better the changes in number and density of cells across development. Each outer square represents a set volume of tissue for each

age group. Neurons are represented by triangles, and nonneurons are represented by circles. The ratio of neurons to nonneurons follows

a distinct developmental trajectory in each brain region. (A) In the neocortex, the total number of cells increases with the overall size of

the structure. However, the number of neurons remains constant from P18 through adulthood. Thus, as the size of the structure increases,

the neuronal density decreases. (B) In the subcortical structures, the total number of cells is lowest at P18, but the total number of neu-

rons is highest at that age, resulting in a high neuronal density. Throughout development, the number of neurons decreases as the total

number of cells increases, resulting in a significantly lowered neuronal density. (C) In the cerebellum, the total number of cells is lowest

at P18, but, because a large proportion of those cells is neuronal, we see a very high neuronal density. The number of neurons remains

constant throughout development into adulthood, but the number of cells (nonneurons) increases, resulting in a lower neuronal density. All

diagrams are based on the total cell density and neuronal density of brain regions at P18, P56, and < P365.



been reported that the vast majority of cortical and sub-

cortical neurons are generated in mice and rats (Dehay

and Kennedy, 2007; Robinson and Dreher, 1990). In rats

cortical neurogenesis begins on E12 and ends on E21,

and in mice cortical neurogenesis begins on E11 and

ends on E19. These dates correspond to E12 and P24,

respectively, in Monodelphis. Thus, the only data from the

current study that we can compare with these previous

studies would be at P35 and progressively later stages.

Probably the most notable difference between the cur-

rent study and these previous investigations is our obser-

vation that the number of cortical neurons stabilizes or

declines following the completion of neurogenesis as

defined in earlier studies (see above), whereas these pre-

vious studies in mice and rats (Bandeira et al., 2009; Lyck

et al., 2007) show a twofold or more increase in the num-

ber of neurons that occurs at P5 in the rat and P16 in the

mouse, well after most studies indicate that neurogenesis

has ended. These studies also demonstrate that this ini-

tial increase in neurons is followed by a reduction in neu-

rons at later postnatal ages. In rats this reduction is as

large as 70%.

Although other authors have reported postnatal neuro-

genesis of GABAergic neurons in mice, the number of

these neurons that actually migrate to the neocortex was

estimated to be relatively modest (Inta et al., 2008). Lyck

and colleagues (2007) suggest that this increase in the

number of neurons in postnatal mice could come from

neurons migrating into the neocortex from other regions

such as the telencephalic wall (Molyneaux et al., 2005;

Noctor et al., 2004) and medial ganglionic eminence

(Anderson et al., 1997; Kriegstein and Noctor, 2004;

Wichterle et al., 2001). Bandiera and colleagues (2009)

ascribe increases in neuronal population to massive post-

natal neurogenesis, a notion counter to all that we know

about neurogenesis in rodents; in fact, their data suggest

that most of neurogenesis occurs postnatally. However,

given the limitation of the methods and the variable effi-

cacy of NeuN for labeling particular populations of neu-

rons and neurons present at early developmental ages

(see above), it is not possible to determine whether the

cells that the authors encountered at these early post-

natal stages are newly born cells, migrating cells or cells

that had not previously expressed NeuN at these earlier

ages. The methods used for the rat suggest but do not

specify that the pyriform cortex was included as part of

the neocortex, which could also account for some of the

differences described (Bandeira et al., 2009).

The cerebellum differs from the other brain regions

described here in that at P18 it makes up only a small

proportion of the weight of the whole brain (5.52%) and

that proportion more than doubles by the time the opos-

sum reaches adulthood (13.15% at P180; Figs. 6, 7, Table

3). Structurally, at P18, the opossum cerebellum is very

immature, consisting of only an external granular layer,

but, by P35, all of the cerebellar layers are apparent,

including the external granular, molecular, and internal

granular layers as well as white matter (Sanchez-Villagra

and Sultan, 2002). This late growth pattern is similar to

that seen in rats, in which the vast majority of cerebellar

growth, including the development of its characteristic

fissures and folia, occurs postnatally (Bandeira et al.,

2009; Carletti and Rossi, 2008; Goldowitz and Hamre,

1998).

As noted above, the number of cerebellar neurons in

Monodelphis throughout all stages of development,

including adults, is low compared with that in eutherian

mammals. This may be due in part to a lack of labeling of

Purkinje cells, lack of labeling of nongranule cell inter-

neurons (at least at earlier stages), lack of efficacy of

labeling cerebellar cells other than the Purkinje cells, or

true species differences. If our small number of cerebel-

lum neurons in adults is due to a lack of Purkinje cell

labeling, and if the ratio of Purkinje cells to granule cells

is similar to that estimated for mice (Goldowitz and

Hamre, 1998; Wetts and Herrup, 1983), then our results

would have underestimated the number of neurons in the

cerebellum by less than 1%.

Cellular composition of the neocortex inother small-brained animals

Another important difference between the current

study and previous studies utilizing similar techniques is

that there are an extremely small number of neurons that

compose the adult marsupial brain compared with esti-

mates from other small-brained mammals such as mice

(Lyck et al., 2007), shrews, and moles (Sarko et al.,

2009). For example, in adult mice, the neocortex contains

14.4 million cells; about 50% are neurons and 50% are

glial cells. Shrews and moles demonstrate a similarity in

the composition of cells within the neocortex; the number

of cells in the small neocortex (0.7 g) of the smoky shrew

was 14 million. As in mice, neurons made up about 50%

of these cells. As the size of the neocortex increased in

shrews and moles, the proportion of neurons to nonneur-

ons changed, with larger brained insectivores having a

greater percentage of nonneurons (e.g., 78% in the hairy-

tailed mole; see Sarko et al., 2009; Table 1), which is simi-

lar to the percentage of nonneurons that we observed in

adult opossums.

The total number of cells in an adult Monodelphis neo-

cortex (0.7 grams) was 3.2 million; 750,000 or 22% were

neurons and 2.5 million (78%) were nonneurons. Thus,

both the total number of cells and the proportions of neu-

rons vs. nonneurons were dramatically different from the



case in rodents and insectivores with a similarly sized neo-

cortex. This suggests that there may be fundamental differ-

ences in signal processing and transmission in marsupial

brains and that glia may play a more important role in infor-

mation processing in marsupials compared with eutherian

mammals (see below). No other studies have examined

the cellular composition of marsupial brains, so it is not

known whether this observation on overall number as well

as proportion of neurons to nonneurons is a general char-

acteristic of marsupials or is specific toMonodelphis.

However, if this were a general feature of marsupials, it

would suggest that early mammals had brains that were

composed of substantially fewer neurons and that glial

cells might have played a more central role in processing.

Changes in the proportion of neurons (increases in num-

ber and density) might have arisen in eutherian mammals

along with more neuron-centered processing networks.

Given the potential role of glial cells in synaptic transmis-

sion (see below), this also suggests that the neocortex of

early mammals might have had a greater capacity for

plastic changes in the adult.

Benefits and limitations the isotropicfractionator method

When considering data generated using the isotropic

fractionator method, it is important to consider both the

benefits of this technique and its limitations. Although

isotropic fractionator methodology does not replace tradi-

tional stereological methods for quantifying various

aspects of neuroanatomical organization and develop-

ment, it offers the extraordinary advantage of estimating

the number and composition of cells in the entire brain or

entire neural structure in a relatively rapid, consistent

manner. Not since the analogous comparative brain mor-

phometry studies of Stephan and colleagues (Baron

et al., 1990; Frahm et al., 1982; Stephan et al., 1981)

have critical and extensive cross-species comparisons

been possible. These early morphometry studies of gross

brain organization and size generated numerous and im-

portant theories regarding brain scaling in mammals (Fin-

lay and Darlington, 1995; Stevens, 2001). Similarly, iso-

tropic fractionator techniques have been used to

compare cellular composition and generate data-driven

theories of cellular evolution and brain scaling in a variety

of mammals, including several primates (Collins et al.,

2010a), different rodents (Herculano-Houzel et al., 2006),

shrews and moles (Sarko et al., 2009), and now marsu-

pials. This accumulation of cross-species data serves as

an important data repository for any number of neuro-

computational, developmental, and evolutionary studies.

Of course, the most obvious limitation of the technique

when used in large structures as a whole is the decon-

struction of tissue to rapidly and accurately estimate the

number of cells, neurons and nonneurons, that make up a

structure. Thus, laminar divisions as well as cortical and

nuclear divisions are lost. Second, there is some destruc-

tion of nuclei as a result of the homogenization process

as well as some loss of nuclei during the immunohisto-

chemical processing of the tissue; however, this loss is

estimated to be minimal (Collins et al., 2010b; Herculano-

Houzel and Lent, 2005). Furthermore, as mentioned

above, there are selected neuron types that NeuN does

not label (Mullen et al., 1992), and the efficacy of NeuN

has not been well characterized in the brains of the many

different animals in which the isotropic fractionator tech-

nique has been used. These limitations must be consid-

ered when interpreting data.

Finally, as we have already mentioned, the isotropic

fractionation method uses dissociated cellular nuclei to

generate data concerning the number and density of cells

within a structure. Because all the cell membranes,

axons, and dendrites have been removed during the cellu-

lar dissociation process, this technique cannot provide

any concrete information about the size, shape, extracel-

lular spaces, or connections of whole cells and neurons.

What about glial cells?The current discussion is based on the assumption that

the nonneuronal cells are predominantly glial cells. The

other types of cells that constitute the nonneuronal

group, endothelial cells, mesothelial cells, and ependymal

cells, are relatively restricted in their distribution. Endo-

thelial cells form the thin lining of blood vessels and com-

pose the blood–brain barrier. The mesothelial cells make

up the pia mater, which in lissencephalic brains is rela-

tively small compared with the volume of tissue that it

encloses. The ependymal cells line the ventricles, whose

membrane volume is substantially smaller than cortical

gray matter. Thus, among the nonneuronal cell types, glial

cells represent the vast majority of this cellular popula-

tion (Morest and Silver, 2003; Temple, 2001).

As noted in the introductory paragraphs, given the

changing role of different glial cells at various stages of de-

velopment and in the adult brain, it is not surprising that

their numbers and density vary across the developmental

time points that we measured as well as in different struc-

tures. Importantly, in adult Monodelphis, the number of

glial cells far exceeds that of neurons. This observation is

particularly important given the present understanding of

glial cells in the adult CNS. Glial cells are no longer consid-

ered to be primarily supportive cells that passively maintain

homeostatic conditions necessary for neurotransmission

but rather actively participate in synaptic transmission. In

recent years, a tripartite synapse, which contains pre- and

postsynaptic neuronal elements as well as astrocytes that



encapsulate the synapse, has been described. Astrocytes

have both ionotropic and metabotropic receptors that

detect neurotransmitters, which increases internal stores

of calcium within the astrocyte. This in turn causes glio-

transmitters to be released at a slower rate than neuro-

transmitters and with a more prolonged affect. This

bidirectional process is thought to modulate neuro-

transmission and plasticity (Pirttimaki and Parri, 2012;

Santello et al., 2012; Verkhratsky et al., 2012). Given

the importance of these cells in both homeostasis and

active synaptic function, it is critical to appreciate the

neuronal/glial cell relationships at a systems level. The

relatively large proportion of glial cells in the adult

Monodelphis suggests that their brains may rely heavily

on glial cells (such as astrocytes) for assisting in the

synaptic transmission of substantially fewer neurons.

This supposition could be explored using the isotropic

fractionator method following the generation of nuclear

markers for different types of glial cells, such as micro-

glia and astrocytes. Ultimately, these studies will lead

to a greater understanding of how neuronal and glial

populations interact and how those interactions may

influence neural processing.

ACKNOWLEDGMENTS

We thank Carol Oxford for her assistance at the UC

Davis Flow Cytometry Shared Resource. We also thank

Cindy Clayton, DVM, and the rest of the animal care staff

at the UC Davis Psychology Department Vivarium.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest.

ROLE OF AUTHORS

All authors had full access to all the data in the study

and take responsibility for the integrity of the data and

the accuracy of the data analysis. Study concept and

design: AMHS, JCD, LAK. Acquisition of data: AMHS.

Analysis and interpretation of data: AMHS. Drafting of the

manuscript: AMHS, LAK. Critical revision of the manu-

script for important intellectual content: AMHS, JCD, LAK.

Statistical analysis: AMHS. Obtained funding: LAK.

Administrative, technical, and material support: JCD.

Study supervision: LAK.

LITERATURE CITEDAitkin L, Nelson J, Farrington M, Swann S. 1991. Neurogene-

sis in the brain auditory pathway of a marsupial, the north-ern native cat (Dasyurus hallucatus). J Comp Neurol 309:250–260.

Anderson SA, Eisenstat DD, Shi L, Rubenstein JL. 1997. Inter-neuron migration from basal forebrain to neocortex: de-pendence on Dlx genes. Science 278:474–476.

Bandeira F, Lent R, Herculano-Houzel S. 2009. Changing num-bers of neuronal and nonneuronal cells underlie postnatalbrain growth in the rat. Proc Natl Acad Sci U S A 106:14108–14113.

Baron G, Stephan H, Frahm HD. 1990. Comparison of brainstructure volumes in Insectivora and primates. IX. Trigemi-nal complex. J Hirnforschung 31:193–200.

Burish MJ, Peebles JK, Baldwin MK, Tavares L, Kaas JH, Hercu-lano-Houzel S. 2010. Cellular scaling rules for primate spi-nal cords. Brain Behav Evol 76:45–59.

Carletti B, Rossi F. 2008. Neurogenesis in the cerebellum.Neuroscientist Rev J Neurobiol Neurol Psychiatry 14:91–100.

Cheung AF, Kondo S, Abdel-Mannan O, Chodroff RA, SireyTM, Bluy LE, Webber N, DeProto J, Karlen SJ, Krubitzer L,Stolp HB, Saunders NR, Molnar Z. 2010. The subventricu-lar zone is the developmental milestone of a 6-layeredneocortex: comparisons in metatherian and eutherianmammals. Cereb Cortex 20:1071–1081.

Collins CE, Airey DC, Young NA, Leitch DB, Kaas JH. 2010a.Neuron densities vary across and within cortical areas inprimates. Proc Natl Acad Sci U S A 107:15927–15932.

Collins CE, Young NA, Flaherty DK, Airey DC, Kaas JH. 2010b.A rapid and reliable method of counting neurons and othercells in brain tissue: a comparison of flow cytometry andmanual counting methods. Front Neuroanat 4:5.

Cunningham CL, Martinez-Cerdeno V, Noctor SC. 2012.Microglia regulate neurogenesis in the developing cerebralcortex. J Neurosci (in press).

Dehay C, Kennedy H. 2007. Cell-cycle control and cortical de-velopment. Nat Rev Neurosci 8:438–450.

Finlay BL, Darlington RB. 1995. Linked regularities in the de-velopment and evolution of mammalian brains. Science268:1578–1584.

Frahm HD, Stephan H, Stephan M. 1982. Comparison of brainstructure volumes in Insectivora and Primates. I. Neocor-tex. J Hirnforschung 23:375–389.

Goldowitz D, Hamre K. 1998. The cells and molecules thatmake a cerebellum. Trends Neurosci 21:375–382.

Herculano-Houzel S, Lent R. 2005. Isotropic fractionator: asimple, rapid method for the quantification of total celland neuron numbers in the brain. J Neurosci 25:2518–2521.

Herculano-Houzel S, Mota B, Lent R. 2006. Cellular scalingrules for rodent brains. Proc Natl Acad Sci U S A 103:12138–12143.

Inta D, Alfonso J, von Engelhardt J, Kreuzberg MM, Meyer AH,van Hooft JA, Monyer H. 2008. Neurogenesis and wide-spread forebrain migration of distinct GABAergic neuronsfrom the postnatal subventricular zone. Proc Natl Acad SciU S A 105:20994–20999.

Kriegstein AR, Noctor SC. 2004. Patterns of neuronal migrationin the embryonic cortex. Trends Neurosci 27:392–399.

Kumar SS, Buckmaster PS. 2007. Neuron-specific nuclearantigen NeuN is not detectable in gerbil subtantia nigrapars reticulata. Brain Res 1142:54–60.

Lyck L, Kroigard T, Finsen B. 2007. Unbiased cell quantifica-tion reveals a continued increase in the number of neo-cortical neurones during early post-natal development inmice. Eur J Neurosci 26:1749–1764.

Marotte LR, Sheng X. 2000. Neurogenesis and identificationof developing layers in the visual cortex of the wallaby(Macropus eugenii). J Comp Neurol 416:131–142.

Martinez-Cerdeno V, Cunningham CL, Camacho J, Antczak JL,Prakash AN, Cziep ME, Walker AI, Noctor SC. 2012. Com-parative analysis of the subventricular zone in rat, ferretand macaque: evidence for an outer subventricular zone inrodents. PloS one 7:e30178.



Miller FD, Gauthier AS. 2007. Timing is everything: makingneurons vs. glia in the developing cortex. Neuron 54:357–369.

Molnar Z, Knott GW, Blakemore C, Saunders NR. 1998. Devel-opment of thalamocortical projections in the South Ameri-can gray short-tailed opossum (Monodelphis domestica). JComp Neurol 398:491–514.

Molyneaux BJ, Arlotta P, Hirata T, Hibi M, Macklis JD. 2005.Fezl is required for the birth and specification of cortico-spinal motor neurons. Neuron 47:817–831.

Morest DK, Silver J. 2003. Precursors of neurons, neuroglia, andependymal cells in the CNS: What are they? Where are theyfrom? How do they get where they are going? Glia 43:6–18.

Mouton PR. 2002. Principles and practices of unbiased ster-eology: an introduction for bioscientists. Baltimore: JohnsHopkins University Press.

Mullen RJ, Buck CR, Smith AM. 1992. NeuN, a neuronal specificnuclear protein in vertebrates. Development 116:201–211.

Nadarajah B, Parnavelas JG. 2002. Modes of neuronal migra-tion in the developing cerebral cortex. Nat Rev Neurosci 3:423–432.

Noctor SC, Martinez-Cerdeno V, Ivic L, Kriegstein AR. 2004.Cortical neurons arise in symmetric and asymmetric divi-sion zones and migrate through specific phases. Nat Neu-rosci 7:136–144.

Noctor SC, Martinez-Cerdeno V, Kriegstein AR. 2008. Distinctbehaviors of neural stem and progenitor cells underliecortical neurogenesis. J Comp Neurol 508:28–44.

Oomman S, Finckbone V, Dertien J, Attridge J, Henne W, MedinaM, Mansouri B, Singh H, Strahlendorf H, Strahlendorf J. 2004.Active caspase-3 expression during postnatal development ofrat cerebellum is not systematically or consistently associatedwith apoptosis. J Comp Neurol 476:154–173.

Pirttimaki TM, Parri HR. 2012. Glutamatergic input–outputproperties of thalamic astrocytes. Neuroscience 205:18–28.

Polazzi E, Contestabile A. 2002. Reciprocal interactionsbetween microglia and neurons: from survival to neuropa-thology. Rev Neurosci 13:221–242.

Puzzolo E, Mallamaci A. 2010. Cortico-cerebral histogenesis inthe opossum Monodelphis domestica: generation of a hexa-laminar neocortex in the absence of a basal proliferativecompartment. Neural Dev 5:8.

Rakic P. 1990. Principles of neural cell migration. Experientia46:882–891.

Robinson SR, Dreher B. 1990. The visual pathways of euther-ian mammals and marsupials develop according to a com-mon timetable. Brain Behav Evol 36:177–195.

Sanchez-Villagra MR, Sultan F. 2002. The cerebellum at birthin therian mammals, with special reference to rodents.Brain Behav Evol 59:101–113.

Sanderson KJ, Weller WL. 1990. Gradients of neurogenesis inpossum neocortex. Brain Res Dev Brain Res 55:269–274.

Santello M, Cali C, Bezzi P. 2012.Gliotransmission and the tri-partite synapse. In:Kreutz MR, Sala C, editors. Synapticplasticity: dynamics, development, and disease. New York:Springer. p 307–331.

Sarko DK, Catania KC, Leitch DB, Kaas JH, Herculano-HouzelS. 2009. Cellular scaling rules of insectivore brains. FrontNeuroanat 3:8.

Sarnat HB, Nochlin D, Born DE. 1998. Neuronal nuclear anti-gen (NeuN): a marker of neuronal maturation in earlyhuman fetal nervous system. Brain Dev 20:88–94.

Saunders NR, Adam E, Reader M, Mollgard K. 1989. Monodel-phis domestica (grey short-tailed opossum): an accessiblemodel for studies of early neocortical development. AnatEmbryol 180:227–236.

Stephan H, Frahm H, Baron G. 1981. New and revised dataon volumes of brain structures in insectivores and prima-tes. Fol Primatol 35:1–29.

Stevens CF. 2001. An evolutionary scaling law for the primatevisual system and its basis in cortical function. Nature411:193–195.

Temple S. 2001. The development of neural stem cells. Na-ture 414:112–117.

Upender MB, Naegele JR. 1999. Activation of microglia duringdevelopmentally regulated cell death in the cerebral cor-tex. Dev Neurosci 21:491–505.

Verkhratsky A, Rodriguez JJ, Parpura V. 2012. Neurotrans-mitters and integration in neuronal–astroglial networks.Neurochem Res (in press).

Wetts R, Herrup K. 1983. Direct correlation between Purkinjeand granule cell number in the cerebella of lurcher chime-ras and wild-type mice. Brain Res 312:41–47.

Weyer A, Schilling K. 2003. Developmental and cell type-spe-cific expression of the neuronal marker NeuN in the mu-rine cerebellum. J Neurosci Res 73:400–409.

Wichterle H, Turnbull DH, Nery S, Fishell G, Alvarez-Buylla A.2001. In utero fate mapping reveals distinct migratorypathways and fates of neurons born in the mammalian ba-sal forebrain. Development 128:3759–3771.

Yan XX, Najbauer J, Woo CC, Dashtipour K, Ribak CE, LeonM. 2001. Expression of active caspase-3 in mitotic andpostmitotic cells of the rat forebrain. J Comp Neurol 433:4–22.



Differential Changes in the Cellular Composition of the Developing Marsupial Brain

Documents